*** lyrrad has quit IRC | 02:35 | |
*** bill_az has quit IRC | 03:49 | |
*** openstackgerrit has quit IRC | 07:47 | |
*** openstackgerrit has joined #swift3 | 07:48 | |
*** acoles_ is now known as acoles | 09:06 | |
*** bill_az has joined #swift3 | 14:33 | |
*** lyrrad has joined #swift3 | 17:13 | |
*** gaul has joined #swift3 | 21:33 | |
*** gaul has left #swift3 | 21:37 | |
*** patchbot` has joined #swift3 | 21:41 | |
*** patchbot has quit IRC | 21:41 | |
*** patchbot` is now known as patchbot | 21:41 | |
*** bill_az_ has joined #swift3 | 21:44 | |
*** bill_az has quit IRC | 21:48 | |
*** acoles is now known as acoles_ | 22:24 | |
kota_ | hello? | 23:00 |
---|---|---|
siva_krishnan | hi | 23:00 |
timburke | hi! | 23:01 |
kota_ | hi all | 23:01 |
kota_ | i think this time is temporary happened the meeting in the last week skipped. | 23:02 |
kota_ | and I have nothing special for today, that's why I didn't update the wiki page. | 23:02 |
kota_ | do you all have something? | 23:03 |
timburke | not really, i'm afraid. i'm still scratching my head over the failures on patch 255069 | 23:04 |
patchbot | timburke: https://review.openstack.org/#/c/255069/ - swift3 - Support x-amz-copy-source-range for Upload Part - ... | 23:04 |
timburke | but that's about it | 23:04 |
kota_ | looks to get jenkins failure | 23:04 |
kota_ | I think that worked well in earlier patch sets. | 23:05 |
timburke | i was thinking that at first, too, but i actually managed to reproduce it on my own VM on occasion. still looking into it; i've got a suspicion it has to do with my re-using an upload id but i'm not sure | 23:06 |
kota_ | oic | 23:09 |
timburke | oh! i did have one other thing to bring up. i don't expect to resolve it now, but wanted to put the problem out there | 23:09 |
siva_krishnan | I have a quick question ? | 23:10 |
siva_krishnan | I have pushed couple of test cases to ceph-s3tests repository . I haven't got any follow up from them. Should I reach out to someone who is handling ceph repo ? | 23:10 |
kota_ | siva_krishnan: sorry, I don't know who guy maintains ceph-s3tests repo. | 23:12 |
timburke | a couple other areas where we could add substantial compatibility are CORS support and Object POST. they seem smaller than ACLs or versioning, but still could be quite useful. however...they both require anonymous access. so... we should probably think about how to enable anonymous S3-like access | 23:12 |
kota_ | timburke: what is you wanted to bring? | 23:13 |
siva_krishnan | timburke noted | 23:13 |
timburke | siva_krishnan: maybe? i'm not entirely sure what their process is like. i can ask alpha_ori at what point we (swiftstack) would be willing to start maintaining a long-lived fork that could incorporate new tests | 23:13 |
siva_krishnan | Ya. that would be better way to go ahead , we can have our own forked version. | 23:15 |
siva_krishnan | I beleive lyrrad has already created a fork of s3test repo | 23:16 |
siva_krishnan | timburke kota_ have question on object POST test cases that have been failing s3compatibility test suite | 23:17 |
lyrrad | I haven’t created one. I just have a reference to a particular commit for the s3 compatibility repo. We could fork in the future if ceph-s3tests isn’t responsive to pulls. | 23:18 |
timburke | siva_krishnan: shoot | 23:19 |
siva_krishnan | we have few test cases faling on Object POST for for bad request paramaeters due to exceptions.AssertionError :: 412 != 400 | 23:19 |
kota_ | looks like we need translateion from 412 (swift precondition failed) to 400 (s3 bad request)? | 23:21 |
siva_krishnan | Shouldn't we use 400 if the request parameters are wrong. I believe 412 should be used for if-* failures | 23:21 |
siva_krishnan | Should it be a change in Swift or Swift3 ? | 23:21 |
kota_ | I think in swift3 if we would change. | 23:22 |
kota_ | but... | 23:22 |
timburke | but to my earlier point, i wonder whether swift3 is doing anything for the request at all | 23:22 |
timburke | (currently) | 23:22 |
kota_ | if-* conditional header would return 3xx when the condition insufficient. | 23:23 |
kota_ | ah, in vaiolation of syntax case? | 23:23 |
timburke | kota_: 304 on GET. 412 on PUT | 23:24 |
timburke | (as i recall) | 23:24 |
kota_ | ah, yeah, Swift doessn't allow if-* conditional except if-Non-Match: * | 23:24 |
kota_ | got it, siva said on *POST* | 23:25 |
siva_krishnan | S3 is returning 412 for PUT | 23:25 |
kota_ | oh, opposite what I was thinking. | 23:25 |
kota_ | no? | 23:26 |
kota_ | so | 23:26 |
kota_ | tim said Swift 412 on PUT and then siva said s3 also 412 on PUT | 23:26 |
siva_krishnan | Swift has been returning 501 for if-* | 23:27 |
siva_krishnan | *PUT | 23:27 |
kota_ | ok, probably I need to look at more deeply | 23:28 |
kota_ | siva_krishnan: could you have launchpad bug report already? | 23:28 |
timburke | hmmm... i do wonder about https://github.com/openstack/swift3/blob/master/swift3/utils.py#L99 though... | 23:28 |
siva_krishnan | Not yet | 23:29 |
kota_ | sounds like it seems good to open the ticket to repro/consider what happen in swift/swift3 and to discuss what we could fix it. | 23:30 |
siva_krishnan | timburke I am saying it based on the test report that was generated after running swift against compatibility test suite | 23:31 |
kota_ | ah, ok, not reproduced by hand yet, right? | 23:31 |
kota_ | it might be a bug on test case? | 23:31 |
siva_krishnan | no. will validate it by hand and let you know | 23:32 |
kota_ | ah | 23:34 |
kota_ | i have an idea for 501 | 23:34 |
kota_ | swift3 doesn't support POST, maybe? and it will return 501 anyways? | 23:34 |
kota_ | siva_krishnan: anyways, I'm looking forward to your report, thanks :) | 23:34 |
kota_ | FYI, https://github.com/openstack/swift3/blob/master/swift3/controllers/obj.py#L117-L118 may trigger 501 | 23:34 |
siva_krishnan | Can that be a case because I thought it is directly communicating to Swift3 with PUT object request | 23:34 |
siva_krishnan | ah ok . | 23:37 |
timburke | Object POST (in S3 -- http://docs.aws.amazon.com/AmazonS3/latest/API/RESTObjectPOST.html) sends POST requests to buckets; swift3 will need to (and does not currently) translate that into a PUT to a new object (along with handling authorization, etc.) | 23:37 |
timburke | controllers/bucket.py has a similar def POST(..): raise S3NotImplemented() | 23:38 |
kota_ | yup, that my mistake to point out controllers/object.py POST though :P | 23:39 |
timburke | oh! but back to why it's a 412! see https://github.com/openstack/swift/blob/45a7a15/swift/proxy/server.py#L358 -- if try something like `curl -v http://saio:8080/asdf`, sure enough, you get back a 412 | 23:41 |
kota_ | wll | 23:43 |
kota_ | well | 23:43 |
siva_krishnan | Should it be changed to 400 ? | 23:44 |
kota_ | i think swift3 cannot handle the 412 because the statement is in out of S3 request right? | 23:45 |
siva_krishnan | yes | 23:45 |
kota_ | looks like client will get 412 when no URL translation from S3 into Swift happen (i.e. no valid Authorization header) | 23:46 |
kota_ | and *IMO* we shouldn't change the response status code according to S3 that different product from OpenStack Swift. | 23:47 |
kota_ | in addition, that response code correctry works in the semantics (i mean precondition failed is a right response when bad url request comming into swift) | 23:48 |
siva_krishnan | ok fine. I just raised this question as I saw return code should be 400 for bad request parameters. Now I understand that request has bad parameters for S3 for swift it bad url request | 23:50 |
timburke | (i'd argue 412 is being abused by swift -- 400 or 404 seem like better fits. that's neither here nor there, though; i'll go bug notmyname about it some other time) | 23:51 |
timburke | so, we're back to needing to be able to (1) recognize that a request should go through the S3 api and (2) decide which account anonymous S3 requests should be routed to | 23:51 |
kota_ | hew | 23:56 |
kota_ | it might be how hard part of swift3 allows anonymous access! | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!